ORIGINAL ARTICLE
Is the “Gök Atlas” Sufficiently Reliable for Forensic Age
Determination of Turkish Children?*
Bora BÜKEN1
Erhan BÜKEN2
Alp Alper ŞAFAK3
Burhan YAZICI3
4
Zerrin ERKOL
Atilla Senih MAYDA5
1
2
3
4
5
Turk J Med Sci
2008; 38 (4): 319-327
© TÜBİTAK
E-mail: [email protected]
Department of Forensic Medicine,
Düzce Faculty of Medicine,
Düzce University,
Düzce - TURKEY
Department of Forensic Medicine,
Faculty of Medicine,
Başkent University,
Ankara - TURKEY
Department of Radiology,
Düzce Faculty of Medicine,
Düzce University,
Düzce - TURKEY
Department of Forensic Medicine,
Faculty of Medicine,
Abant İzzet Baysal University,
Bolu - TURKEY
Department of Public Health,
Düzce Faculty of Medicine,
Düzce University,
Düzce - TURKEY
Received: September 21, 2007
Accepted: May 05, 2008
Aim: We investigated whether the contemporary use of the Gök Atlas method is sufficiently reliable for
forensic age estimations of Turkish children.
Materials and Methods: Plain radiographs of 248 females and 298 males between 11 and 22 years of age
were taken. Paired sample t-tests, Pearson correlation coefficients, and linear regression were used to
determine the differences and model the relationships between mean skeletal (bone) age (BA) and mean
chronological age (CA), using SPSS 11.5 statistical software.
Results: In all cases, the CA and BA scores were significantly different and showed high correlation (P < 0.05).
The regression model was significant (P < 0.01). According to age groups, the BA was greater than the CA
for all age groups by 0.09–3.10 years for boys and 0.44-3.05 years for girls, and this difference was
significant for all age groups >14 years for boys and for those 11-18 years of age for girls. The standard
deviation of the difference between BA and CA for boys between 11 and 18 years of age and for girls between
11 and 20 years of age was >1 year.
Conclusions: We found a large discrepancy between CA and BA according to the Gök Atlas. This discrepancy
may be significantly misleading, especially in criminal liability cases.
Key Words: Age estimation, bone age, Gök Atlas method, skeletal maturation, Turkish children
Günümüz Türk Çocuklarında Adli Tibbi Yaş Tayini İçin “GÖK Atlası”nın
Kullanımı Yeterince Güvenli mi?
Amaç: Bu çalışmada GÖK atlas metodunun adli tıbbi yaş tayininde, günümüz Türk çocukları için yeterli olup
olmadığı araştırmaktadır.
Yöntem ve Gereç: 11-22 yaş arasında 298 erkek 248 kız, sağlıklı çocuğun planlı radyografileri çekildi.
Ortalama kronolojik yaş (CA) ve GÖK Atlas metoduna göre elde edilen ortalama iskelet yaşı (BA) arasında ki
farkları saptamak ve ilişkiyi modellemek için paired sample t test, pearson korelasyon and regresyon analizi
kullanıldı. Analizler SPSS 11.5 ortamında gerçekleştirildi.
Bulgular: CA ve BA karşılaştırıldı. Çalışmaya katılan tüm olgular yaş gruplarına ayrılmaksızın ele alındığında
her iki cinsiyette fark istatistiksel olarak anlamlı idi (P < 0.05). Yüksek korelasyon vardı. Determinasyon
katsaysı (R²) erkeklerde 0.861, kızlarda 0.625’e eşitti. Regresyon modeli anlamlı bulundu (P < 0.01). Fakat
yaş gruplarına göre, Tüm yaş gruplarında erkeklerde BA (0.09-3.10 yıl arasında) ileri ve 14 yaşın üstünde
tüm çocuklarda fark istatistiksel olarak anlamlı, kızlarda BA (0.44 - 3.05 yıl) ileri ve 11-18 yaş arasında kız
çocuklarda fark istatistiksel olarak anlamlı idi. Erkeklerde 11-18 , kızlarda 11-20 yaş arasında tüm yaş
gruplarında farkların standart sapması 1 yıldan daha fazla idi.
Sonuç: Biz CA ve GÖK atlasına göre saptanan BA arasında yaş gruplarına gore her iki cinsiyette büyük bir
farklılık saptadık. Bu farklığın özellikle ceza sorumluluğuyla ilgili olarak yaş tayini yapılan olgularda önemli hak
kayıplarına neden olabileceği düşünülmektedir.
Correspondence
Bora BÜKEN
Department of Forensic Medicine
Düzce Faculty of Medicine
Düzce University
Konuralp 81620
Düzce - TURKEY
Anahtar Sözcükler: Yaş tayini, Kemik yaşı, İskelet gelişimi, Türk çocuklar, GÖK atlas metodu
[email protected]
* This study was supported by the Abant İzzet Baysal University research fund.
319
BÜKEN, B et al.
Is the Gök Atlas Reliable for Identifying Age in Turkish Children?
Introduction
Age determination is one of the most important
factors for human identification. Age estimation studies
indicate that identification procedures are not only useful
in solving personal or social problems, but can also have
international ramifications. At the Institute of Legal
Medicine of Berlin University, 7% of the forensic age
assessments were of Turkish people (1). Age
determinations have been requested from courts of law
for Turks living in various foreign countries, particularly
in Europe. Incorrect evaluations can cause severe legal
problems. When appropriate criteria are not used in
forensic age determinations, a significant loss of rights
can occur in sentencing and in legal judgments (2-4).
Minors are subject to special Turkish national and
community regulations that are different from those
applied to people over 18 years of age. According to
these regulations, minors under 12 years of age are
exempt from criminal liability, whereas minors between
12 and 15 and between 15 and 18 years of age are
subject to special age-related criminal standards.
According to civil codes, an individual begins adulthood
when he or she attains 18 years of age. Boys and girls
cannot marry until they have reached 18 years of age,
although under extenuating circumstances, a judge can
give permission for boys and girls of 16 years to marry.
In addition, in the penal code, any kind of sexual behavior
is defined as exploitation when it involves children who
are younger than 16 years. If the contents of the law are
taken into consideration, the age categories of younger
than 12, 15, and 18 years affect many aspects of the law
from sentencing to civil rights (5,6).
In Turkey, the evaluation of age for legal purposes is
conducted by pediatricians, radiologists, and forensic
medicine specialists. A survey of forensic medicine
specialists indicated that 45.7% use the Gök Atlas,
21.7% use the Greulich-Pyle (GP) method, and 17.4%
rely on the Tanner-Whitehouse method (7). Very few
studies have evaluated the appropriateness of these
methods for use in Turkish children (8,9). Despite its
widespread use, we found no published studies of
whether the Gök Atlas is appropriate for use in the age
determination of Turkish children.
Numerous international studies have reported that
racial differences affect growth and development and that
over time, differences in growth and development also
320
Turk J Med Sci
occur in communities in which studies have been
conducted. This necessitates making adjustments in the
values that are currently used in age evaluations (10,11).
Therefore, our purpose was to investigate whether the
atlas published by Gök et al. (2), which is widely used in
Turkey, is reliable for contemporary use in Turkish
children. Here, we present our results for Turkish boys
and girls.
Materials and Methods
Twelve schools in Central Düzce, Turkey, were
sampled randomly. A total of 1200 children (boys and
girls) who were born and registered in local hospitals
were randomly chosen from these schools. Informed
consent was obtained from the parents of 772 children
who were then included in our study. Children from 11
to 22 years of age with no past history of chronic or
severe systemic illness and without previous history of
trauma or injury at the measured joints and bones were
selected for study. Height was measured to the nearest
0.1 cm using a height scale, and weight was measured to
the nearest 0.1 kg using an electronic weighting scale
(SECA, Hamburg, Germany). The children were between
the 3rd and 97th percentiles according to height and
weight standards determined by Neyzi et al. (12). The
staging system developed by Tanner was used to assess
the patterns of penile and testicular features, pubic hair
in the boys, and breast development, pubic hair in girls
(13,14).
All children were examined by a physician to exclude
the possibility of any diseases that might have affected
their physical growth. The study group comprised 546
subjects who met the study criteria: 248 (45.4%)
females and 298 (54.6%) males (12,15). A questionnaire
was completed by the parents of each child to obtain
demographic data. The same physician conducted genital
examinations and measurements, and the study director
reexamined some of the subjects to maintain
standardization and reliability.
Conventional roentgenograms of left hands and
wrists, shoulder, elbow, and pelvis were taken. The
roentgenographic examinations were done by Trophy
radiograph using green-sensitive, 18 × 24, 24 × 30, and
30 × 40 film (Kodak). Exposure doses were calculated
according to age, zone of exposure, and tissue thickness.
The exposed doses ranged from 46 to 80 kV and 6.5 to
Vol: 38
Is the Gök Atlas Reliable for Identifying Age in Turkish Children?
No: 4
25 mAs. All radiographs were compared with the Gök
Atlas by an independent forensic medicine specialist who
is currently responsible for legal cases involving age
determination. During the evaluation, the physician
(forensic specialist) was informed only of the sex of the
child. The concordance between two readings of the
researcher (forensic specialist) was evaluated by
computing the intraclass correlation coefficient (ICC). The
ICC for boys was r = 0.970 (n = 51) and for girls was r
= 0.999 (n = 60).
August 2008
Results
Boys
The mean chronological age (CA) of subjects was
compared to the mean skeletal (bone) age (BA) according
to the Gök Atlas using a paired t-test, and their
correlation was evaluated using the Pearson correlation
coefficient. Linear regression was used to model the
relationship between CA and BA according to the Gök
Atlas. The subjects were then divided into yearly age
groups, and the differences between the CA and BA were
determined for each age group, as in previous studies (4).
The statistical analyses were conducted using SPSS 11.5
software.
The mean ± SD chronological age (CA) of boys was
16.17 ± 3.11 years (SE: 0.15, median: 15.93, range:
11.13–23.76 years), whereas the mean ± SD skeletal age
(BA) was 17.49 ± 3.87 years (SE: 0.18, median: 19.00,
range: 10.00–22.00 years; Table 1a). The difference
between CA and BA was 1.32 ± 1.98 and statistically
significant (P < 0.001). The standard deviation of BA was
greater than the standard deviation of CA. BA was lower
than the CA in 21 boys (7.0%), equal to the CA in 65
boys (21.8%), and greater than the CA in 212 boys
(71.1%). The measured growth factors were highly
correlated with CA and BA (Table 2a). There was a high
correlation (Pearson r = 0.861; P < 0.001) between the
mean CA and the mean BA (Table 2a), but according to
age groups, significant correlations occurred for boys
aged 12 and 13 (Table 3a). For boys between 11 and 22
years of age, the mean BA was greater than the mean CA
for all ages except 22-year-olds, and was significantly
greater in boys aged 15 and over.
This study was approved by the Hospital Ethics
Committee of Abant Izzet Baysal University, Turkey, and
was supported by a research fund from the university.
Epiphyseal union was complete for 10.3%, 15.2%,
33.3%, 41.2%, 39.1%, 66.7%, 93.8%, and 100% of
boys aged 15, 16, 17, 18, 19, 20, 21, and 22,
Table 1a. Mean age and anthropometric characteristics of Turkish boys
(n = 298).
Table 1b. Mean age and anthropometric characteristics of Turkish girls
(n = 248).
Characteristic
Mean ± SD
Median
Characteristic
16.17 ± 3.11
15.93
Chronological age
Skeletal age
17.49 ± 3.87
19.00
Weight (kg)
60.74 ± 17.94
59.4
Height (cm)
164.22 ± 14.39
167.2
Chronological age
Table 2a. Correlation of chronological age and skeletal age with sexual
maturation criteria for Turkish boys.
Chronological age
(CA)
Skeletal age
(BA)
Mean ± SD
Median
14.89 ± 2.55
14.71
Skeletal age
17.77 ± 3.04
18.00
Weight (kg)
51.27 ± 10.02
52.5
Height (cm)
154.94 ± 8.66
156.1
Table 2b. Correlation of chronological age and skeletal age with sexual
maturation criteria for girls according to Tanner.
Chronological age
Skeletal age
Characteristic
Characteristic
r
p
r
p
0.691
0.000
0.745
0.000
r
p
r
p
Pubic hair
0.730
0.000
0.861
0.000
Breast development
0.549
0.000
0.614
0.000
Genital development
0.726
0.000
0.806
0.000
Menarche
0.626
0.000
0.716
0.000
Skeletal age
0.861
0.000
1
–
Skeletal age
0.791
0.000
1
–
Pubic hair
321
Is the Gök Atlas Reliable for Identifying Age in Turkish Children?
BÜKEN, B et al.
Turk J Med Sci
Table 3a. Mean ± SD chronological age (CA) and skeletal age (BA) of Turkish boys and differences between CA and BA according
to age (paired t-test).
Correlation
Mean BA
(Gök)
Difference
CI: 95%
t
11.47 ± 0.25
11.84 ± 0.99
–0.37 ± 1.09
–1.69
26
12.45 ± 0.26
12.80 ± 1.47
–0.35 ± 1.34
34
13.46 ± 0.30
13.56 ± 1.52
–0.09 ± 1.41
14
39
14.57 ± 0.24
15.15 ± 2.43
15
29
15.52 ± 0.31
18.62 ± 2.33
16
33
16.50 ± 0.29
19.15 ± 1.75
–2.64 ± 1.73
–8.76
0.000
0.139
0.442
17
27
17.40 ± 0.32
20.48 ± 1.39
–3.08 ± 1.39
–11.51
0.000
–0.135
0.501
18
17
18.46 ± 0.30
20.71 ± 1.53
–2.23 ± 1.52
–6.07
0.000
0.137
0.600
19
23
19.43 ± 0.29
20.82 ± 1.02
–1.39 ± 1.00
–6.67
0.000
–0.237
0.277
20
21
20.54 ± 0.31
21.33 ± 1.01
–0.79 ± 0.25
–3.20
0.004
–0.222
0.334
21
16
21.48 ± 0.29
21.93 ± 0.25
–0.45 ± 0.32
–5.61
0.000
0.280
0.294
22
8
22.62 ± 0.29
22.00 ± 0.00
0.62 ± 0.29
6.15
0.000
–
–
Age
groups
N
11
25
12
13
Mean CA
P value
r
p
0.104
–0.295
0.153
–1.36
0.186
0.557
0.003
–0.39
0.699
0.446
0.008
–0.58 ± 2.39
–1.51
0.138
0.232
0.155
–3.10 ± 2.38
–6.98
0.000
–0.099
0.608
Table 3b. Mean ± SD chronological age (CA) and skeletal age (BA) of Turkish girls and differences between CA and BA according to
age (paired t-test).
Correlation
Age
groups
N
11
39
12
35
13
14
Mean CA
Difference
CI: 95%
11.52 ± 0.28
13.43 ± 1.62
-1.91 ± 1.57
- 7.60
0.000
0.252
0.121
12.57 ± 0.30
15.42 ± 1.69
-2.85 ± 1.64
- 10.23
0.000
0.212
0.221
29
13.50 ± 0.28
16.55 ± 2.01
-3.05 ± 2.14
- 8.23
0.000
0.132
0.496
35
14.58 ± 0.27
18.68 ± 1.59
-4.11 ± 1.51
-16.07
0.000
0.351
0.038
15
24
15.43 ± 0.34
18.67 ± 1.09
-3.24 ± 1.17
-13.58
0.000
-0.082
0.705
16
31
16.46 ± 0.29
20.32 ± 1.62
-3.86 ± 1.62
-13.26
0.000
-0.086
0.645
17
23
17.50 ± 0.36
20.34 ± 1.66
-2.84 ± 1.62
- 8.38
0.000
0.225
0.302
18
16
18.27 ± 0.23
20.31 ± 1.82
-2.03 ± 1.82
- 4.45
0.000
-0.004
0.988
19
8
19.54 ± 0.20
20.50 ± 1.60
-0.95 ± 1.75
- 1.54
0.167
- 0.725
0.042
20
4
20.55 ± 0.08
21.00 ± 1.41
-0.44 ± 1.45
- 0.69
0.580
-0.403
0.597
21
4
21.51 ± 0.25
22.00 ± 0.00
-0.49 ± 0.25
- 3.90
0.030
-
-
respectively. Epiphyseal union was complete for all boys
over 21 years of age.
The coefficient of determination (R ) was 0.861,
indicating that 86.1% of the variation in Gök (BA) can be
explained by CA. The regression model, tested by analysis
of variance (ANOVA), was significant (P < 0.01). The
322
t
Mean BA
(Gök)
P value
r
p
intercept of the regression was 0.156 and was significant
(P > 0.05). The slope of the regression was 1.072 and
was also significant (P < 0.01). For boys, the Gök Atlas
method has a 0.156-year backward shift for the model
and every year of CA increases the Gök-based estimation
by a rate of 1.072 (Figure 1a).
Vol: 38
Is the Gök Atlas Reliable for Identifying Age in Turkish Children?
No: 4
August 2008
Figure 1a. Regression model of chronological age (CA) and skeletal age
(BA) according to the Gök Atlas for Turkish boys (R2 =
0.861).
Figure 1b. Regression model for girls, R2 = 0.625. CA: chronological
age; BA: skeletal age according to Gök Atlas.
We also examined the distribution of sexual
maturation criteria by age (Table 4a). The family incomes
were less than minimum wage for 45.5% of boys and
greater than the average income for 6.4% of boys. The
boys’ nutrition was based on diets consisting of
vegetables and foods made with flour.
old, respectively. For 15-year-olds, there were no cases
of complete epiphyseal union, but epiphyseal union was
complete in all subjects over 20 years of age.
Girls
The mean CA of girls was 14.89 ± 2.55 SD years (SE:
0.12, median: 14.71, range: 11.06 - 21.77 years),
whereas mean BA according to the Gök Atlas was
significantly higher, at 17.77 ± 3.04 SD years (SE: 0.11,
median: 18.00, range: 11 - 21 years; P < 0.001; Table
1b). BA was less than the CA in only 2 girls (0.8%), equal
in 15 (6.0%), and greater than the CA in 231 (93.1%).
There was a high correlation (Pearson r = 0.791; P <
0.001) between CA and BA for all girls, but when
analyzed according to age group, significant correlations
were found only at 14 years of age (Tables 2b-3b). The
growth factors showed high correlations with both CA
and BA (Table 2b).
CA and BA of girls were compared according to age
group (Table 3b). In the age groups of girls between 11
and 21 years old, BA was advanced compared to CA for
all age groups. P values were statistically significant for all
ages except 19- and 20-year-old subjects.
Radiographs showed complete epiphyseal union in
3.4%, 8.6%, 0%, 41.9%, 47.8%, 43.8%, 50.0%, and
50.0% of girls in each age group from 13 to 20 years
To model the relationships between BA and CA, the
linear regression method was used, where CA and BA
were the independent and dependent variables,
respectively. The determination coefficient (R ), which
represents the fraction of variability in y that can be
explained by the variability in x, was 0.625; thus,
0.625% of Gök (BA) variation could be explained by CA.
The regression model was also tested by ANOVA and was
found to be significant (P < 0.01). The intercept and
slope of the regression model were 3.769 and 0.942,
respectively, and both were significant (P < 0.01). Thus,
for girls, the Gök Atlas method has a 3.7-year forward
shift for the model, and every year of CA increases the
Gök-based estimation at a rate of 0.942 (Fig. 1b).
The percentage of no menarche was 27.8% for all
subjects, 87.2% at 11 years, 60.0% at 12 years, 41.4%
at 13 years, and 5.7% at 14 years; all girls older than 14
years had menarche.
The family incomes of 39.5% of the girls were below
minimum wage, whereas those of only 3.6% were above
average.
We also examined the distribution of sexual
maturation criteria (Table 4b).
The subjects’ nutrition was based on diets consisting
mainly of vegetables and foods made with flour.
323
Is the Gök Atlas Reliable for Identifying Age in Turkish Children?
BÜKEN, B et al.
Turk J Med Sci
Table 4a. Distribution of subjects according to Tanner’s sexual development criteria for boys.
Stage 1
Stage 2
Stage 3
Stage 4
Stage 5
Criterion
n
%
n
%
n
%
n
%
n
%
Pubic hair
37
12.4
36
12.1
38
12.8
153
51.3
34
11.4
Penis and testis development
38
12.8
35
11.7
38
12.8
154
51.7
33
11.1
Table 4b. Distribution of subjects according to Tanner sexual development criteria for girls.
Stage 1
Stage 2
Stage 3
Stage 4
Stage 5
n
%
n
%
n
%
N
%
n
%
PH
25
10.1
31
12.5
68
27.4
103
41.5
21
8.5
BD
18
7.3
55
22.2
71
28.6
89
35.9
15
6.0
PH: pubic hair; BD: breast development.
Discussion
Physicians are often requested to assess the age of
an individual in civil and criminal cases. The study of the
epiphyseal union of bones is considered a reasonable
scientific and accepted method of age estimation by
courts of law worldwide (16). Skeletal age
determination is usually performed by comparing the
plain radiograph of a patient with findings in a normal
reference population (17,18). The Gök Atlas method is
the most commonly used by forensic specialists in
determining skeletal maturity from X-rays in Turkey (7).
Using the Gök Atlas, the skeletal ages of boys between
the ages of 11 and 22 were determined by examining
the degree of epiphyseal fusion of shoulder, elbow, hand
and wrist, and pelvic bones (2). Our objective was to
determine whether the Gök Atlas is sufficiently accurate
for estimating the age of Turkish boys and girls for
forensic purposes.
It should be kept in mind that clear medico-legal
statements about age cannot be made from radiographs
alone (19). Of the forensic methods recommended for
age estimation, the evaluation of sexual maturity shows
the largest range of variation and therefore should be
used for age determination only in conjunction with an
evaluation of skeletal maturity (20). Age estimations
324
should involve careful examination of anthropometric
data, signs of sexual maturation, and potential agerelevant developmental disorders (19). Therefore, we
assessed the degree of sexual maturation, age, and
anthropometric characteristics of all of the subjects in our
study (Table 1).
According to DIE (Republic of Turkey, Turkish
Statistical Institute), the average monthly income for a
family in Düzce in 2001 was 1242 YTL (approximately
$750 U.S.) (21). The family incomes were less than
minimum wage for 45.5% of boys and above the average
income for Duzce for 6.4% of boys. The monthly
minimum wage in Turkey was 480 YTL (approximately
$350 U.S. dollars) in 2005 (20). Thus, the boys in our
study population had low to moderate socioeconomic
status. Although extreme situations of poverty and
malnourishment can lead to pronounced delays in skeletal
maturation (17), our study was designed to only include
subjects with age-related height and weight between the
3rd and 97th percentile of a normal population. These
standards were determined by Neyzi et al. (12) for
Turkish children. Therefore, the subjects could be reliably
assumed to be developmentally normal. Thus, poverty
likely had little influence on our results and was not a
cause of undernourishment in our study population.
Vol: 38
No: 4
Is the Gök Atlas Reliable for Identifying Age in Turkish Children?
Although the Gök Atlas is used by approximately 50%
of forensic medicine specialists (7), no studies have been
published on its relevance since its publication. Therefore,
we were unable to compare our results with others.
We found significant differences between CA and BA
for the entire group of boys and girls (P < 0.001).
According to age groups, the mean BA was greater than
the mean CA by 0.09–3.10 years for all ages groups and
was significantly greater for boys older than 14 years
(Table 3a). Mean BA was advanced (0.44-3.86 years) for
all ages, and differences were significant at all ages except
for 19-20 years for girls (Table 3b). Methods for age
estimation accepted by courts of law worldwide offer
reliability ranges. For example, Groell et al. (17) showed
that there is a discrepancy of up to 1 year between CA
and measured BA and confirmed that the GP method may
reliably be applied to Central European children.
However, there is no previously determined reliability
range for the Gök Atlas method. In particular, we found
differences >1 year for boys between 15 and 19 years of
age and for all girls between 11 and 18 years of age. It
was >3 years between 13 and 16 years of age. The
differences between BA and CA gradually increased
between 11 and 14 years of age for girls. We think that
a suitable discrepancy range should not exceed 1 year;
thus, the discrepancy for the Gök Atlas method is
unsuitable for forensic age estimations. For this reason,
we recommend that this method not be used for age
estimations of early post-pubertal boys and of girls
between the ages of 11 and 18.
The difference between CA and BA was <1 year
(0.09–0.58 years) for boys between 11 and 14 years and
was not statistically significant. However, the difference
between CA and BA was >1 year for boys between 14
and 19 years and was statistically significant (Table 3a).
This greater difference may be explained by puberty.
According to Rikhasor et al. (22), the age limit for
skeletal growth retardation is 15 years for boys, and this
age is approximately at or around the time of puberty for
boys. In our study, boys’ skeletal maturation accelerated
after 14 years of age.
The difference between the BA and CA was <1 year
(0.44 - 0.95 years) for girls at the ages of 19, 20, and
21, but that of the other age groups was >1 year (Table
3b). We are uncertain about the reliability for subjects
older than 19 years in our study, because the epiphyseal
union was completed after age 19 in >50% of our cases,
August 2008
and the number of cases in which it was not complete in
these age groups was insufficient for a complete analysis.
According to Rikhasor et al. (22), the age limit of
skeletal retardation in Pakistani girls was 13 years. The
age of puberty for Turkish girls was similar according to
Akarsu et al. (23), whereas in our study, all girls older
than 14 years had menarche and were skeletally mature.
The distribution of sexual maturation criteria with
skeletal development and the correlation between CA and
sexual maturation criteria are shown in Tables 2b and 4b.
We are uncertain about the reliability of the Gök Atlas
method for boys older than 20 years because the
epiphyseal union was complete by the age of 20 for
>50% of boys. The overall correlations were high (i.e.,
boys of all ages combined; Table 2a), but this may be
misleading because correlations for the separate ages
were very low and were significant only for boys 12 and
13 years of age (Table 3a) and average only for 14-yearold girls (Tables 2a-b and 3a-b). For this reason, this
method does not appear to be technically useful for
forensic age estimation in Turkish girls and boys.
Greulich and Pyle (24) identified standard deviations
ranging from 0.6 to 1.1 years for their method, but no
range was given for the Gök Atlas method. If we accept
these criteria as Gök Atlas criteria, we would not be able
to use this method for boys between 11 and 19 years old
and especially for those of 14 and 15 years and for girls
between 11 and 20 years old.
Many recent studies have indicated a comparative
acceleration of skeletal maturity in Western and other
populations. The cause of this acceleration could be new
sociocultural factors rather than economic conditions
(23,24). The Gök Atlas was last published in 1985, and
its initial publication was in 1969. Gök et al. (2), in the
atlas preface, state that the atlas can be used for Turkish
children, but there have been major changes in the
Turkish population since 1969 caused both by economic
and sociocultural factors. This large discrepancy in CA and
BA could have arisen from these causes.
Our results are very important for Turkish children
who live in Turkey and in foreign countries. Age
determinations have been requested from the courts for
some Turks living in various foreign countries,
particularly in Europe. Thus, incorrect evaluations could
cause legal problems to arise outside the country. For this
reason, nationwide studies need to be conducted as soon
325
BÜKEN, B et al.
Is the Gök Atlas Reliable for Identifying Age in Turkish Children?
as possible to determine an appropriate atlas for age
estimation in Turkish children. As a short-term solution,
it is necessary to determine the appropriateness of
internationally accepted atlases for Turkish children and
to make adaptations to atlases for Turkish children.
Our study has several limitations. The children that
were enrolled in our study were all from Düzce province;
thus, differences that may be related to geographic
location or climate were minimized. Racial and climatic
differences are very high among geographical regions in
Turkey. Therefore, we recruited our study population
only from the Black Sea region of Turkey. However, this
study should be repeated in other regions of Turkey, and
we hope that our results will encourage other forensic
researchers to do so.
We could not determine hormone levels (e.g., FSH,
LH, testosterone) because of economic limitations and
ethical considerations. We could not determine tooth
development because we did not have orthopantography
equipment; this equipment is not available in our hospital
or in our province. We planned the study according to
Neyzi’s standards (12), which reflect Turkish children,
but these standards do not include adults.
An advantage of our study is that it was designed to
avoid the effects of growth retardation and obesity on
skeletal maturation by using only children with agerelated height and weight between the 3rd and 97th
percentiles for the general population. In addition, we
determined the degree of sexual maturation of all
candidates to account for the confounding effect of
puberty on skeletal maturation. Lastly, bone
determination was analyzed by a forensic medicine
specialist who is very experienced in the practice of age
estimation to minimize subjectivity and interobserver
Turk J Med Sci
variability. All cases were evaluated twice. When the
analysis results were discordant, these cases were reevaluated and the results were registered as decisive. This
was necessary to produce good results. The strong
concordance in evaluations is a reflection of proficiency
and confirms indications that bone age estimations
improve with clinical experience (4).
A Final Note
To our knowledge, this type of study has not been
conducted previously for this atlas. We found a large
discrepancy between BA and CA for all cases combined
and in separate ages for both sexes. This discrepancy may
be misleading, especially in criminal liability cases. This is
an important human rights problem for Turkish children.
This atlas should not be used particularly for subjects
between 15 and 18 years of age for forensic age
estimations in criminal liability cases.
We found less discrepancy between CA and BA for
Turkish boys using the same statistical method in
comparable cases in a previous study (4). Therefore, we
suggest that the Greulich-Pyle Atlas should be used
instead of the Gök Atlas for boys and girls, particularly
those between 11 and 18 years of age.
Acknowledgements
This study was supported by the Abant Izzet Baysal
University research fund. A portion of this study was
presented as a poster presentation at the Fifth National
Forensic Science Congress, Samsun, Turkey. We thank
the Düzce Ministry of Education and Prof. Dr. Kenan
Kocabay for their support and Dr. Filiz Demir and Dr. Ali
Koyuncuer for their assistance in the measurement phase
of this study.
References
1.
Ritz-Timme S, Cattaneo C, Collins MJ, Waite ER, Schütz HW. Age
estimation: the state of the art in relation to the specific demands
of forensic practise. Int J Leg Med 2000; 113: 129-36.
2.
Gök Ş, Erölçer N, Özen C. Age determination in forensic medicine.
nd
2 ed. Istanbul: Turkish Republic Ministry of Justice, Council of
Forensic Medicine Press; 1985. pp. 1-26. [in Turkish]
3.
326
Özen HC, Kırangil B, Fincancı ŞK. Determination of identity. Adli
Tıp Dergisi 1988; 4: 173-81. [in Turkish]
4.
Buken B, Safak AA, Yazici B, Buken E, Mayda AS. Is the
assessment of bone age by the Greulich-Pyle method reliable at
forensic age estimation for Turkish children? Forensic Sci Int
2007; 173: 146-53.
5.
Turkish Penal Code. Law no.: 5237. Accepted date: 26
September 2004. Official Gazette 2004; (25611) (in Turkish).
6.
Turkish Civil Code. Law no.: 4721, Accepted date: 22 November
2001. Official Gazette 8.12.2001, (24607) (in Turkish).
Vol: 38
7.
8.
9.
No: 4
Is the Gök Atlas Reliable for Identifying Age in Turkish Children?
Büken B, Büken E, Yazıcı B, Şafak A, Mayda A. Is the Use of the
Gök Atlas Sufficiently Reliable for Age Determination of Turkish
Children Today? Fifth Annual Anatolian Forensic Science
Congress, 8–10 September 2006, Samsun, Turkey. 2006; 41.
Koç A, Karaoğlanoğlu M, Erdoğan M, Kösecik M, Cesur Y.
Assessment of bone ages: Is the Greulich- Pyle method sufficient
for Turkish boys? Pediatr Int 2001; 43: 662-5.
Ersoy UO. Cross-sectional study of the evaluation of bone age in
the 0–18-year-old population. Thesis Specialist. Turkish Republic
Ministry of Justice, Council of Forensic Medicine. İstanbul, 2003.
[in Turkish]
10.
Scheming A, Reisinger W, Loreck D, Vendura K, Markus W,
Geserick G. Effects of ethnicity on skeletal maturation:
consequences for forensic age estimations. Int J Leg Med 2000;
113: 253-8.
11.
Scanderbeg AC, Sacco MC, Gialloreti LE, Fraracci L. Skeletal age
assessment in children and young adults: comparison between a
newly developed sonographic method and conventional methods.
Skeletal Radiol 1998; 27: 271-7.
August 2008
15.
Lisa LY. Correlation of sexual maturation with skeletal age of
southern Chinese girls. Aust Orthod J 1997; 14(4): 215-7.
16.
Banerjee KK, Agarwal BBL. Estimation of age from epiphyseal
union at the wrist and ankle joints in the capital city of India.
Forensic Sci Int 1998; 1–2: 31-9.
17.
Groell R, Lindbichler F, Riepl T, Gherra L, Roposch A, Fotter R.
The reliability of bone age determination in central European
children using the Greulich and Pyle method. Br J Radiol 1999;
72: 461-4.
18.
Lewis CP, Lavy CBD, Harrison WJ. Delay in skeletal maturity in
Malawian children. J Bone Joint Surg Br 2002; 84-B: 732-4.
19.
Milner GR, Levick RK, Kay R. Assessment of bone age: a
comparison of the Greulich and Pyle, and the Tanner and
Whitehouse methods. Clin Radiol 1986; 7: 119-21.
20.
Schmeling A, Reisinger W, Geserick G, Olze A. Age estimation of
unaccompanied minors Part I. General considerations. Forensic
Sci Int 2006; 159: 61-4.
21.
Düzce City Development Project. 2004; 95. [in Turkish]
12.
Neyzi O, Ertuğrul T. Pediatrics 1. Nobel Tıp Kitapevi Press; 1986.
p. 62-81. [in Turkish]
22.
Rikhasor RM, Qureshi AM, Rathi SL, Channa NA. Skeletal
maturity in Pakistani Children. J Anat 1999; 195: 305-8.
13.
Büyükgebiz A, Eroğlu Y, Karaman O, Kinik E. Height and weight
measurements of male Turkish adolescents according to biological
maturation. Acta Pediatr Jpn 1994; 36: 80-3.
23.
Akarsu S, Aygün AD, Güvenç H, Kocabay K. Pubertal development
of eastern Turkish girls in Elazığ. T Klin J Med Res 1998; 16: 169.
14.
Mora S, Boechat MI, Pietka E, Huang HK, Gilsanz V. Skeletal age
determinations in children of European and African descent:
applicability of the Greulich and Pyle standards. Pediatr Res 2001;
50(5): 624-8.
24.
Greulich WW, Pyle SI. Radiographic atlas of skeletal development
of hand and wrist. California: Stanford University Press; 1959.
327